Ansible role with a reboot fails

Archives Ansible Project
1

I’ve trying to write an Ansible role that upgrades Github. The command to upgrade the server “ghe-upgrade github-2.13.1.pkg” for example, reboots the server, and when that happens, I get, the following connection because my ssh connection drops of course.

`

TASK [github : Upgrade github1.company.com with version 2.13.1] **********
fatal: [github1.company.com]: UNREACHABLE! => {“changed”: false, “msg”: "Failed to connect to the host via ssh: ", “unreachable”: true}
to retry, use: --limit @/build/workspace/team_SCSCM/github/github-upgrade.retry

`

I tried the ideas from ansible-reboot-and-wait.yml exmaple, and others with similar approach, to no avail. So my tasks look like

`

This command will reboot the node

  • name: Upgrade Github host
    command: “ghe-upgrade {{ pkg_dir }}/{{ pkg_name }} --yes”

(It craps out after above task)

Now we will run a local ‘ansible -m ping’ on this host until it returns.

This works with the existing ansible hosts inventory and so any custom ansible_ssh_hosts definitions are being used

  • local_action: shell ansible -u {{ ansible_ssh_user }} -m ping github1.company.com
    register: result
    until: result.rc == 0
    retries: 30
    delay: 10

`

Any pointers on how I can make this work? Thanks!

2

Sorry, ignore my last post… I can see that is exactly what you did

3

I tried the ideas from ansible-reboot-and-wait.yml exmaple
<https://gist.github.com/infernix/a968f23c4f4e1d6723e4&gt;, and others with

That's over 4 years old, in this profession that is a lifetime.

similar approach, to no avail. So my tasks look like

# This command will reboot the node
- name: Upgrade Github host
  command: "ghe-upgrade {{ pkg_dir }}/{{ pkg_name }} --yes"

(It craps out after above task)

# Now we will run a local 'ansible -m ping' on this host until it returns.
# This works with the existing ansible hosts inventory and so any custom
ansible_ssh_hosts definitions are being used
- local_action: shell ansible -u {{ ansible_ssh_user }} -m ping
github1.company.com
  register: result
  until: result.rc == 0
  retries: 30
  delay: 10

Any pointers on how I can make this work? Thanks!

If it's possible to make ghe-upgrade not reboot the server that would be easiest I guess.
The reason is you don't not how long the upgrade process will take and make it hard to find a value for the delay bellow.

The reboot is as simple as

- name: reboot server
   shell: sleep 2 && reboot
   async: 1
   poll: 0

- name: Wait for server
   wait_for_connection:
     delay: 60

The delay must be at least as long it takes for the server to shutdown ssh, it not it will be able to connect again and the play continues.

4

Not quite exactly - there was no pause task in the OP’s version.

The linked webpage says that the pause is needed to prevent the existing control connection being used.

Regards, K.

5

Thanks all. Yes there should’ve been a pause between the two tasks. And yes there is an option to not reboot during the ghe-upgrade procedure, but I still would need to reboot twice after the command, once for Github to finish its upgrade, and another for the upgrade to take effect. So I still would to use the tricks mentioned here.